Research map / Vision-language and multimodal / Human and embodied modeling
PET2Rep: Towards Vision-Language Model-Drived Automated Radiology Report Generation for Positron Emission Tomography
Yichi Zhang, Wenbo Zhang, Zehui Ling, Gang Feng, Sisi Peng, Deshu Chen, Yuchen Liu, Hongwei Zhang, Shuqi Wang, Lanlan Li, Limei Han, Yuan Cheng and 3 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| evaluate VLMs against a simple template baseline | state-of-the-art VLMs · PET2Rep | Overall (%): 27.5 vs 26.8 (Lingshu-32B fused), 26.8 (MedGemma-4B) | Hurt |
| use fused PET/CT image instead of separate PET and CT images (ablation) | MedGemma-27B · PET2Rep | Overall (%): 17.1 -> 20.1 | Improved |
| use fused PET/CT image instead of separate PET and CT images (ablation) | HuatuoGPT-Vision · PET2Rep | Overall (%): 14.0 -> 23.0 | Improved |
| use fused PET/CT image instead of separate PET and CT images (ablation) | LLaVa-V1.5-7B · PET2Rep | Overall (%): 14.6 -> 9.2 | Hurt |
| use fused PET/CT image instead of separate PET and CT images (ablation) | Qwen2.5-VL-7B · PET2Rep | Overall (%): 26.4 -> 26.6 | No clear effect |
| use fused PET/CT image instead of separate PET and CT images (ablation) | Lingshu-32B · PET2Rep | Overall (%): 26.7 -> 26.8 | No clear effect |
| scale model size within a family | Qwen2.5-VL (7B vs 32B vs 72B) · PET2Rep | Overall (%): 26.4 -> 17.6 -> 17.3 (separate inputs) | Hurt |
| scale InternVL3 from 8B to 78B | InternVL3 · PET2Rep | Overall (%): 24.3 -> 21.7 (separate inputs) | Hurt |
| use medical-specific VLMs rather than general-purpose | VLMs · PET2Rep | Overall (%): 26.8 (MedGemma-4B) vs 26.6 (Qwen2.5-VL-7B); 16.2 (MedVLM-R1) | Mixed |
Closest papers by meaning
- Toward a Vision-Language Foundation Model for Medical Data: Multimodal Dataset and Benchmarks for Vietnamese PET/CT Report Generation (NeurIPS 2025 · similarity 0.89)
- S2D-Align: Shallow-to-Deep Auxiliary Learning for Anatomically-Grounded Radiology Report Generation (AAAI 2026 · similarity 0.85)
- LLM-RG4: Flexible and Factual Radiology Report Generation Across Diverse Input Contexts (AAAI 2025 · similarity 0.84)
- Large-scale and Fine-grained Vision-language Pre-training for Enhanced CT Image Understanding (ICLR 2025 · similarity 0.84)
- HC-LLM: Historical-Constrained Large Language Models for Radiology Report Generation (AAAI 2025 · similarity 0.84)
- G2D: From Global to Dense Radiography Representation Learning via Vision-Language Pre-training (NeurIPS 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
